Biomarker for predicting success rate of embryo implantation and application thereof

Abstract

The invention discloses a biomarker for predicting the success rate of embryo implantation and application of the biomarker. The biomarker comprises the following components: hsa-miR-877 to 3p, has-miR-125a to 5p, has-miR-22 to 3p, has-miR-363 to 3p, has-miR-122 to 5p, hsa-let-7a to 5p, and hsa-miR-10a to 3p. The invention also discloses a method for preparing the biomarker. According to the method, the optimal feature combination is obtained by performing correlation analysis and screening on plasma miRNA of patients with successful embryo implantation and failed embryo implantation, and the prediction model is established by utilizing a machine learning method, so that the success rate of embryo implantation can be effectively predicted, and the problem that in the prior art, when embryo implantation outcome is predicted, the success rate of embryo implantation cannot be effectively predicted is effectively solved. The sample extraction process is complex, invasive and invasive, and the accuracy is remarkably improved.

Description

Technical Field

[0001] The invention belongs to the field of biomedical technology and relates to a biomarker for predicting embryo implantation success rate and its application. Background Art

[0002] In vitro fertilization-embryo transfer (IVF-ET) has become an effective way for infertile women to achieve pregnancy as soon as possible. Although the development and technological innovation of assisted reproductive technology have enabled most infertile patients to obtain high-quality embryos and then become pregnant through in vitro embryo culture technology, some high-quality embryos are still lost in the early or even very early stages of pregnancy. Studies have shown that there is a certain correlation between embryonic factors and endometrial receptivity factors and successful embryo implantation. Being able to predict whether the embryo can be successfully implanted by evaluating the maternal condition before embryo transfer can not only improve pregnancy outcomes, but also avoid the waste of high-quality embryos.

[0003] Endometrial receptivity refers to the state of molecular and cellular transformation of the endometrium to promote embryo attachment, growth and invasion. The window of implantation (WOI) is a short and specific time window in the endometrial cycle. At this time, the endometrium is receptive and can accept the attachment, growth and invasion of the blastocyst. It usually occurs between the 19th and 24th days of the menstrual cycle, which is the 7th day after the peak of luteinizing hormone (LH) in the natural cycle (LH+7) or the 5th day after progesterone administration in the artificial cycle (P+5). However, not all women have the same WOI. Some women have WOI displacement, including early or delayed, and some women have shortened WOI, which makes the dialogue between the embryo and the endometrium out of sync, leading to implantation failure. Therefore, accurate assessment of the endometrial receptivity and implantation window is of great significance for the successful implantation of the embryo.

[0004] The currently published techniques for predicting RIF mainly include: detecting and evaluating endometrial receptivity through endometrial biopsy, such as CN110724736A discloses an IVF-ET-related microRNA and its application, specifically including a kit for preparing a method for determining the pregnancy outcome of in vitro fertilization and embryo transfer (IVF-ET), wherein the miRNA is selected from the following group: hsa-miR-27a-3p, hsa-miR-127-3p, hsa-miR-29a-3p, hsa-miR-100-5p, hsa-miR-486-5p, hsa-miR-23a-3p or a combination thereof. However, the current methods are usually invasive and invasive, and the process of biopsy of endometrial tissue will also cause changes in the endometrial environment, which has limitations in clinical use.

[0005] In summary, the development of biomarkers for predicting the success rate of embryo implantation can non-invasively and accurately predict the success rate of embryo implantation, which is of great significance in the field of embryo transplantation. Summary of the invention

[0006] In view of the deficiencies of the prior art and actual needs, the present invention provides a biomarker for predicting the success rate of embryo implantation and its application, screens maternal plasma miRNA markers, and constructs a detection model to predict the outcome of embryo transplantation based on maternal serum, which is of great significance to the field of embryo transplantation.

[0007] To achieve the above object, the present invention adopts the following technical solutions:

[0008] In a first aspect, the present invention provides a biomarker for predicting the success rate of embryo implantation, wherein the biomarker includes: hsa-miR-877-3p, has-miR-125a-5p, has-miR-22-3p, has-miR-363-3p, has-miR-122-5p, hsa-let-7a-5p and hsa-miR-10a-3p.

[0009] The present invention conducts correlation analysis and screening of plasma miRNAs in patients with successful and failed embryo implantation, and develops a miRNA marker combination based on maternal plasma, which can be effectively used to predict the success rate of embryo implantation. This effectively solves the problem of the prior art that the sample extraction process is complex, invasive and invasive when predicting the outcome of embryo implantation, and the accuracy is significantly improved.

[0010] In the present invention, the nucleic acid sequences of the biomarkers are shown in Table 1.

[0011] Table 1

[0012] Markers Nucleic acid sequence serial number hsa-miR-877-3p UCCUCUUCUCCCUCCUCCCAG SEQ ID NO.1 hsa-miR-125a-5p UCCCUGAGACCCUUUAACCUGUGA SEQ ID NO.2 hsa-miR-22-3p AAAAUGGUUCCCUUUAGAGUGU SEQ ID NO.3 hsa-miR-363-3p AAUUGCACGGUAUCCAUCUGUA SEQ ID NO.4 hsa-miR-122-5p UGGAGUGUGACAAUGGUGUUUG SEQ ID NO.5 hsa-let-7a-5p UGAGGUAGUAGGUUGUAUAGUU SEQ ID NO.6 hsa-miR-10a-3p CAAAUUCGUAUCUAGGGGAAUA SEQ ID NO.7

[0013] In a second aspect, the present invention provides the use of the biomarker for predicting the success rate of embryo implantation and / or its detection reagent described in the first aspect in the preparation of a product for predicting the success rate of embryo implantation.

[0014] In a third aspect, the present invention provides a kit for predicting embryo implantation success rate, the kit comprising a reagent for detecting the expression level of the biomarker for predicting embryo implantation success rate described in the first aspect.

[0015] In a fourth aspect, the present invention provides the use of the biomarker for predicting the success rate of embryo implantation and the detection reagent thereof as described in the first aspect in predicting the success rate of embryo implantation for the purpose of non-disease diagnosis.

[0016] In a fifth aspect, the present invention provides a method for constructing a model for predicting embryo implantation success rate, the method comprising:

[0017] (1) determining the expression level x of the biomarker for predicting embryo implantation success rate according to claim 1 in the sample;

[0018] (2) The characteristic expression is normalized and standardized using equations (1) and (2), wherein x' is the normalized and standardized characteristic expression, μ is the mean of the characteristic expression, σ is the variance, and x max is the maximum value of feature expression, x min The minimum value of feature expression was used to construct a model for detecting endometrial receptivity using machine learning methods;

[0019]

[0020]

[0021] In the present invention, the methods commonly used in the art for detecting miRNA expression are applicable to the present invention. For example, in a specific embodiment of the present invention, the miRNA expression can be detected by sequencing technology, which may include the following steps:

[0022] 1) Filter the raw data of plasma miRNA sequencing: remove 3' end adapters, sequences without adapter insertion, sequences containing polyA and polyT, sequences with only AT or CG, sequences with a single base (A / T / C / G) accounting for more than 80% of a sequence, and sequences with a length of less than 18 nt after removing adapters; and remove duplicates from the filtered sequences;

[0023] 2) Align the filtered sequences to the Rfam database, obtain annotations for the aligned sequences, and extract the sequences aligned to miRNA and those not aligned as the next valid data;

[0024] 3) Quantify the valid data and identify known miRNAs.

[0025] In the present invention, the markers are screened by constructing a training set, comprising the following steps:

[0026] 1) For each miRNA, its expression level is required to be higher than 20 in 70% of the samples, otherwise it will be removed from the feature pool

[0027] 2) Calculate the correlation between different miRNA expression profiles, requiring that the correlation between any two miRNA expression profiles be less than 0.8;

[0028] 3) Use the Pearson correlation coefficient, chi-square test, and F test between features and labels to screen out preliminary feature combinations;

[0029] 4) Use the logistic regression model for pre-training to obtain the feature importance coefficient of the model, and then select the optimal feature combination based on the coefficient.

[0030] Preferably, the sample comprises any one of serum, plasma, uterine fluid, uterine lavage fluid, vaginal exfoliated cells, vaginal secretions or endometrial biopsy products, or a combination of at least two thereof.

[0031] Preferably, step (1) further comprises the step of standardizing the expression level according to formula (3).

[0032] miRNA i RPM =miRNA i reads / total miRNA reads ×10 6 Formula (3).

[0033] Among them, miRNA i reads For each miRNA expression level, miRNA i RPMs The results after normalization of each miRNA expression level, total miRNA reads It is the sum of all miRNA expression levels.

[0034] Preferably, the machine learning method in step (2) includes logistic regression, random forest, support vector machine or generalized linear model.

[0035] Preferably, the code for constructing a model for predicting embryo implantation success rate is:

[0036]

[0037] 'wb');

[0038] The custom functions are as follows:

[0039] def split_and_transform(train_data,train_label,rand_state):

[0040] x_train,x_test,y_train,y_test=train_test_split(train_data,train_label,test_size=0.2,random_state=rand_state)

[0041] from sklearn.preprocessing import StandardScaler,MinMaxScaler

[0042] mm,st=MinMaxScaler(),StandardScaler()

[0043] mm.fit(x_train)

[0044] x_train,x_test=mm.transform(x_train),mm.transform(x_test)

[0045] st.fit(x_train)

[0046] x_train,x_test=st.transform(x_train),st.transform(x_test)

[0047] return x_train,x_test,y_train,y_test,mm,st

[0048] def create_new_model_pred(mode='lr'):

[0049] if mode == 'lr':

[0050] model=LogisticRegression(random_state=0,class_weight='balanced')

[0051] return model

[0052] elif mode == 'svm':

[0053] model=SVC(probability=True)

[0054] return model

[0055] elif mode == 'nb':

[0056] model=GaussianNB()

[0057] return model

[0058] elif mode == 'rf':

[0059] model=RandomForestClassifier(n_estimators=10)

[0060] return model

[0061] else:

[0062] raise ValueError().

[0063] In a sixth aspect, the present invention provides a model for predicting the success rate of embryo implantation, wherein the model for predicting the success rate of embryo implantation is constructed by the method for constructing a model for predicting the success rate of embryo implantation described in the fifth aspect.

[0064] The present invention conducts correlation analysis and screening of plasma miRNA of patients with successful and failed embryo implantation, develops a group of miRNA markers related to successful embryo implantation, and establishes a prediction model using machine learning methods to accurately predict the success rate of embryo implantation.

[0065] In a seventh aspect, the present invention provides a method for predicting embryo implantation success rate for the purpose of non-disease diagnosis, the method comprising:

[0066] A female sample to be tested is taken, and the expression level of the biomarker for predicting the success rate of embryo implantation described in the first aspect in the sample is determined, and the expression level is input into the model for predicting the success rate of embryo implantation described in the sixth aspect to predict the success rate of embryo implantation.

[0067] In the present invention, the sample may include any one of serum, plasma, uterine fluid, uterine lavage fluid, vaginal exfoliated cells, vaginal secretions or endometrial biopsy products, or a combination of at least two of them.

[0068] Compared with the prior art, the present invention has the following beneficial effects:

[0069] The present invention provides a group of plasma miRNA markers related to successful embryo implantation, which mainly solves the problem that the sample extraction process is complex, invasive and invasive when predicting whether the embryo can successfully implant. At the same time, a model for predicting the success rate of embryo implantation is constructed based on a machine learning method, which can accurately predict the success rate of embryo implantation based on maternal serum, significantly improve the prediction accuracy, and the area under the receiver operating characteristic curve (ROC) (Area Under the Curve, AUC) of the effect of predicting successful embryo implantation can reach 0.88. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 Analyze the flow chart for the present invention;

[0071] Figure 2 The effect diagram of predicting embryo implantation success rate in the training set constructed by the present invention;

[0072] Figure 3 This is a graph showing the effect of the method constructed in the present invention on predicting the success rate of embryo implantation in the test set. DETAILED DESCRIPTION

[0073] To further illustrate the technical means and effects of the present invention, the present invention is further described below in conjunction with the embodiments and drawings. It should be understood that the specific implementation methods described herein are only used to explain the present invention, rather than to limit the present invention.

[0074] If no specific techniques or conditions are specified in the examples, the techniques or conditions described in the literature in the field or the product instructions are used. If no manufacturer is specified for the reagents or instruments used, they are all conventional products that can be purchased through regular channels.

[0075] In a specific embodiment of the present invention, the plasma miRNA of patients with successful and failed embryo implantation is correlated and screened to obtain the optimal feature combination, and a prediction model is established using a machine learning method, which can accurately predict the success rate of embryo transplantation based on maternal serum. The complete analysis flow chart is as follows: Figure 1 shown.

[0076] Example 1

[0077] This example constructs a model for predicting the success rate of embryo implantation.

[0078] 1. Obtaining plasma samples

[0079] In this example, peripheral blood samples on the ovulation day of 45 patients with successful embryo transfer and 15 patients with failed embryo transfer were used for a control study, and high-throughput sequencing was performed on peripheral blood plasma miRNA of the two samples, with the number of original sequencing sequences for each sample being greater than 20M.

[0080] 2. Analysis of plasma miRNA expression

[0081] (1) The raw data of sample miRNA sequencing were filtered: 3' end adapters, sequences without adapter insertion, sequences containing polyA and polyT, sequences with only AT or CG, sequences with a single base (A / T / C / G) accounting for more than 80% of a sequence, and sequences with a length of less than 18 nt after removing the adapter; and the filtered sequences were deduplicated.

[0082] (2) Align the filtered sequences to the Rfam database, obtain annotations for the aligned sequences, and select the sequences that are aligned to miRNA and those that may not be aligned as valid data. The software parameters used are the default parameters.

[0083] (3) Quantify the valid data and identify known miRNAs.

[0084] (4) The known miRNA quantitative results were standardized according to the following formula:

[0085] miRNA i RPM = miRNA i reads / total miRNA reads ×10 6 Formula (3);

[0086] Among them, miRNA i reads For each miRNA expression level, miRNA i RPMs The results after normalization of each miRNA expression level, total miRNA reads It is the sum of all miRNA expression levels.

[0087] 3. Screening of features to predict embryo transfer outcomes

[0088] (1) For the plasma miRNA expression data obtained above, based on the expression of each miRNA in the sample, 70% of the miRNAs with expression levels below 20 in the samples were screened out.

[0089] (2) The correlation between different miRNA expression profiles was calculated, requiring that the correlation between any two miRNA expression profiles be less than 0.8, that is, one of the two miRNAs with a correlation higher than 0.8 would be randomly eliminated.

[0090] (3) The Pearson correlation coefficient between features and labels, chi-square test and F test are used to screen the most important features. In this embodiment, the first 10 features of each method are taken and the union is calculated to screen out the preliminary feature combination.

[0091] (4) Pre-training was performed using a logistic regression model to obtain the feature importance coefficient of the model. In this embodiment, a feature set with a feature importance coefficient greater than 0.5 and less than -0.5 was taken as a candidate feature combination, and finally 7 miRNAs (including hsa-miR-877-3p, has-miR-125a-5p, has-miR-22-3p, has-miR-363-3p, has-miR-122-5p, hsa-let-7a-5p, and hsa-miR-10a-3p) were included in the construction of the prediction model.

[0092] 4. Construction of a model to predict the success rate of embryo implantation.

[0093] (1) Normalize and standardize the selected optimal feature set.

[0094]

[0095] (2) Plasma miRNA samples from 45 patients with successful embryo transfer and 15 patients with failed embryo transfer were selected as the training set. In each round of training, the samples in the training set were randomly split, 80% of which were used for model training and 20% for model validation. A total of 50 rounds of training were performed to eliminate the impact of sample randomness.

[0096] (3) Use machine learning methods to build a prediction model based on the above-mentioned confirmed features in the training set.

[0097] (4) Calculate the AUC of the model on the training set. The results are as follows Figure 2As shown in the figure, in the training set, the AUC of the model built using logistic regression (LR) can reach 0.97, the AUC of the model built using support vector machine (SVM) can also reach 0.97, the AUC of the model built by naive Bayes (NB) can reach 0.94, and the AUC of the model built using random forest (RF) can reach 1.0.

[0098] The code for building a model to predict the success rate of embryo implantation is:

[0099]

[0100]

[0101] The custom functions are as follows:

[0102]

[0103] x_train,x_test=mm.transform(x_train),mm.transform(x_test)

[0104] st.fit(x_train)

[0105] x_train,x_test=st.transform(x_train),st.transform(x_test)

[0106] return x_train,x_test,y_train,y_test,mm,st

[0107] def create_new_model_pred(mode='lr'):

[0108] if mode == 'lr':

[0109] model=LogisticRegression(random_state=0,class_weight='balanced')

[0110] return model

[0111] elif mode == 'svm':

[0112] model=SVC(probability=True)

[0113] return model

[0114] elif mode == 'nb':

[0115] model=GaussianNB()

[0116] return model

[0117] elif mode == 'rf':

[0118] model=RandomForestClassifier(n_estimators=10)

[0119] return model

[0120] else:

[0121] raise ValueError().

[0122] Example 2

[0123] In this example, the effect of the model constructed in Example 1 was verified in plasma miRNA samples of 18 patients with successful embryo transplantation and plasma miRNA samples on the ovulation day of 6 patients with failed embryo transplantation.

[0124] 1. The steps of obtaining plasma samples and analyzing the expression of plasma miRNA are the same as those in Example 1.

[0125] 2. Extract the 7 miRNA features obtained in Example 1 for each sample, normalize the features of the test set samples with reference to Example 1, and use the model constructed in Example 1 as input to predict the probability of successful embryo implantation of the sample.

[0126] 3. The computational model predicts the effectiveness of embryo transfer outcomes in the test set.

[0127] Calculate the AUC of the model on the test set, the results are as follows Figure 3 As shown, we can see that on the test set, the AUC of the model built using logistic regression (LR) can reach 0.88, the AUC of the model built using support vector machines (SVM) can also reach 0.8, the AUC of the model built by naive Bayes (NB) can reach 0.83, and the AUC of the model built using random forest (RF) reaches 0.88.

[0128] In summary, the present invention provides a group of miRNA markers related to successful embryo implantation, which mainly solves the problem that the sample extraction process is complex, invasive and invasive when predicting whether the embryo can successfully implant. A prediction model is established by using a machine learning method. The area under the receiver operating characteristic curve (ROC) of the final model for predicting successful embryo implantation reaches 0.88, and the accuracy of the prediction is significantly improved.

[0129] The applicant declares that the present invention illustrates the detailed method of the present invention through the above-mentioned embodiments, but the present invention is not limited to the above-mentioned detailed method, that is, it does not mean that the present invention must rely on the above-mentioned detailed method to be implemented. Those skilled in the art should understand that any improvement of the present invention, equivalent replacement of various raw materials of the product of the present invention, addition of auxiliary components, selection of specific methods, etc., all fall within the protection scope and disclosure scope of the present invention.

Claims

1. A biomarker for predicting embryo implantation success rate, characterized in that: The biomarkers include: hsa-miR-877-3p, has-miR-125a-5p, has-miR-22-3p, has-miR-363-3p, has-miR-122-5p, hsa-let-7a-5p and hsa-miR-10a-3p.

2. Use of the biomarker for predicting embryo implantation success rate and / or its detection reagent as described in claim 1 in the preparation of a product for predicting embryo implantation success rate.

3. A kit for predicting embryo implantation success rate, characterized in that: The kit includes reagents for detecting the expression level of the biomarker for predicting the success rate of embryo implantation as claimed in claim 1.

4. Use of the biomarker for predicting embryo implantation success rate as described in claim 1 in predicting embryo implantation success rate for non-disease diagnosis purposes.

5. A method for constructing a model for predicting embryo implantation success rate, characterized in that: The method comprises: (1) determining the expression level x of the biomarker for predicting embryo implantation success rate according to claim 1 in the sample; (2) The expression level is normalized and standardized using equations (1) and (2), wherein x' is the normalized and standardized expression level, μ is the mean of the expression level, σ is the variance, and x max is the maximum expression value, x min The minimum value of feature expression was used to construct a model for predicting embryo implantation success rate using machine learning methods; 6. The method for constructing a model for predicting embryo implantation success rate according to claim 5, characterized in that: The sample includes any one of serum, plasma, uterine fluid, uterine lavage fluid, vaginal exfoliated cells, vaginal secretions or endometrial biopsy products, or a combination of at least two of them.

7. The method for constructing a model for predicting embryo implantation success rate according to claim 5 or 6, characterized in that: Step (1) also includes the step of normalizing the expression amount according to formula (3); miRNA iRPM = miRNA ireads / total miRNA reads ×10 6 Equation (3); Among them, miRNA ireads For each miRNA expression level, miRNA iRPMs The results after normalization of each miRNA expression level, total miRNA reads It is the sum of all miRNA expression levels; Preferably, the machine learning method in step (2) includes logistic regression, random forest, support vector machine or generalized linear model.

8. A model for predicting embryo implantation success rate, characterized in that: The model for predicting the success rate of embryo implantation is constructed by the method for constructing a model for predicting the success rate of embryo implantation described in any one of claims 5-7.

9. A method for predicting embryo implantation success rate for non-disease diagnosis purposes, characterized in that: The method comprises: A female sample to be tested is taken, and the expression level of the biomarker for predicting the success rate of embryo implantation as described in claim 1 in the sample is measured, and the expression level is input into the model for predicting the success rate of embryo implantation as described in claim 8 to predict the success rate of embryo implantation.

10. The method for predicting embryo implantation success rate for the purpose of non-disease diagnosis according to claim 9, characterized in that: The sample includes any one of serum, plasma, uterine fluid, uterine lavage fluid, vaginal exfoliated cells, vaginal secretions or endometrial biopsy products, or a combination of at least two of them.

Images